Force Multiple For Loops to run Sequentially

Question

I have the following sequence of 3 for loops, and I need the last for loop to execute after the other two have properly incremented the *in pointer, and that they are through executing before the data begins processing in the third loop. The purpose of the 1st 2 loops is just to skip over header data.
#include <math.h>
#include <stdint.h>
#include "xilly_debug.h"
#define SKIPBYTES 896
#define SKIPCOUNT 448 // (896/2)
// When creating data cube, it is helpful to discard unneeded bands.
#define SHORTEST_BAND_TO_INCLUDE 171 // Must be between 1 and 519
#define SHORTEST_SKIPCOUNT 119016 // (171*696*2/2)
#define LONGEST_BAND_TO_INCLUDE 453 // Must be between 2 and 520, greater than SHORTEST_BAND_TO_INCLUDE
#define LONGEST_SKIPCOUNT 46632 // (67*696*2/2)
#define PROCESSCOUNT 196968 // (283*696*2/2)
// BANDS = LONGEST_BAND_TO_INCLUDE - SHORTEST_BAND_TO_INCLUDE + 1;
#define BANDS 283
#define RADIANCE_SCALING_FACTOR 100
uint16_t radCorr(uint16_t unCorrPixel, int coeffIndex){
uint16_t corrPixel; //Holder for the corrected pixel value
float a2, b2, c2; //Intermediate variables for the actual calculations
if (coeffIndex > 282)
coeffIndex = 0;
c2 = cf[coeffIndex];
a2 = (float) unCorrPixel;// * cf_mn;//L0R
b2 = a2*c2*RADIANCE_SCALING_FACTOR;
corrPixel = *((uint16_t *) &b2); // Convert float to uint16_t
return corrPixel;
void xillybus_wrapper(int *in, int *out) {
#pragma AP interface ap_fifo port=in
#pragma AP interface ap_fifo port=out
#pragma AP interface ap_ctrl_none port=return
uint32_t a1; //Holds input data
uint16_t x1; //Holds input after converting to float; input to function
uint16_t y1; //HOlds output of function, to return to processor
static int n = 0; //Index for correlation matrix
static int i = 0;
// Handle input data
//Skip over SKIPBYTES
for(i = 0; i < SKIPCOUNT; i++){
*in++;
if(i<20){
xilly_puts("Skipcount Index Value:\n");
xilly_decprint(n, 5);
xilly_puts("\n");
xilly_puts("Value of *in pointer:\n");
xilly_decprint(*in, 5);
xilly_puts("\n");
//Discard Lower Bands
for(i = 0; i < SHORTEST_BAND_TO_INCLUDE; i++){
*in++;
if(i<20){
xilly_puts("Shortest Band to Include Index Value:\n");
xilly_decprint(n, 5);
xilly_puts("\n");
xilly_puts("Value of *in pointer:\n");
xilly_decprint(*in, 5);
xilly_puts("\n");
n++;
for (i = 0; i < PROCESSCOUNT; i++) {
a1 = *in++;
x1 = *((uint16_t *) &a1); // Convert uint32_t to uint16_t
if(i<20){
xilly_puts("Processing Loop Iteration (i) Value\n");
xilly_hexprint(i, 5);
xilly_puts("\n");
xilly_puts("Input Value To radCorr Function:\n");
xilly_hexprint(x1, 5);
xilly_puts("\n");
xilly_puts("Index Value Before Function:\n");
xilly_decprint(n, 5);
xilly_puts("\n");
xilly_puts("Value of *in pointer:\n");
xilly_decprint(*in, 5);
xilly_puts("\n");
// xilly_puts("Coefficient Value Function:\n");
// xilly_decprint(cf[n], 5);
// xilly_puts("\n");
// Run the calculations
y1 = radCorr(x1, n++);
if(i<20){
xilly_puts("Index Value after Function:\n");
xilly_decprint(n, 5);
xilly_puts("\n");
// Handle output data
xilly_puts("Returned Value from Function:\n");
xilly_hexprint(y1, 5);
xilly_puts("\n");
xilly_puts("\n");
*out++ = y1;

Answer 1

all loops are accessing a shared resource (*in++) with read consequences. It would be difficult to imagine HLS would schedule these in parallel. If you want to make sure, I guess you can add a ap_wait() between the loops but I am pretty sure this is not necessary.